On Friday 05 September 2008 20:32:32 Yuri Timofeev wrote: > 2008/9/5 Kern Sibbald <[EMAIL PROTECTED]>: > >From what I understand, it *does* work, but just slowly. It is not correct > > to say that nothing is done. It has been working and doing its job it > > may have even been removing orphaned records in this time. > > No. This is not a work. > > Mysql all the time (3 days) performs the same request: > > SELECT File.FileId, Job.JobId FROM File LEFT OUTER JOIN Job ON > (File.JobId = Job.JobId) WHERE Job.JobId IS NULL LIMIT 300000; > > Orphaned records are not deleted. > There was a difference between 3 days and 3 minutes.
How big is your database? How many file records? It sounds to me like you have something very badly tuned or you have a badly broken MySQL. > > dbcheck does not solve the problem of disk space and you as a > developer, it should be well known. > > > I don't see that there was any need for a reply. You know the solution > > for you. Create the index, or write a script that > > creates it, runs dbcheck and then removes it. > > When I first encountered a problem, I do not wrote. > And the problems not only to me. See bacula-users. > > > If someone submits a patch to dbcheck that improves it and makes it > > automatic, all the better. > > So I made the desired patch. > > And you asked: whether to embed the code dbcheck.c code, which depends > on a specific database. > > I do not ask you to do the work. I ask the council. Maybe later I can > still do some useful patch. > But I need to start something. We explained our general philosophy -- it is a generality not anything absolute. If you want a specific answer for specific code, you will need to send the code. > > > This is an Open Source program. You know your solution -- create an > > index. > > I made the desired patch. > > And I asked you: whether to embed the code dbcheck.c code, which > depends on a specific database? > Since in the other database - (probably?) all OK. > > I do not ask you to do the work. I ask the council. > Maybe later I can still do some another useful patch for bacula. That would be nice. > But I need to start something. That is good. > > We can go for the easiest way: include an appropriate comment in the > file src/cats/make_mysql_tables.in after rows > CREATE TABLE File... They already have a similar comment. Yes, I have your patch for that and will include it in the near future. > > Such pacht I have already sent. > It is hard to do? > > 2008/9/5 Kern Sibbald <[EMAIL PROTECTED]>: > > Hello, > > > > On Thursday 04 September 2008 20:24:29 Yuri Timofeev wrote: > >> Hi > >> > >> Some time ago I received a problem - dbcheck does not work with options: > > > > From what I understand, it *does* work, but just slowly. > > > >> 8) Check for orphaned File records > >> 9) Check for orphaned Path records > >> > >> dbcheck worked 3 days and nothing is done. I stopped dbcheck (ctrl+c) > >> and restart MySQL. > > > > It is not correct to say that nothing is done. It has been working and > > doing its job it may have even been removing orphaned records in this > > time. > > > >> I explain what I need dbcheck: > >> > >> If you do not have enough space on the disk, the bacula (all versions) > >> *very* bad handles this situation. > > > > If you do not have enough disk space on your system, practically every > > program that exists will have problems, and Bacula does nee a lot of > > storage. > > > >> And this is another big problem. With the shortage of disk space > >> bacula could damageŠ² Volumes. > >> I saw it many times. > > > > Running dbcheck is not going to resolve a disk shortage problem. It may > > free up space in your catalog, but that space will remain unless you > > explicitly compact your catalog as documented in the manual. > > > >> So I had to check the database Catalog for orphaned File records and > >> orphaned Path records. > > > > In my opinion, there are better ways of freeing up space, including > > compacting your database. > > > >> The time by pass and in maillist bacula-users described the same > >> situation: > >> > >> [Bacula-users] > >> From: Ryan Novosielski <[EMAIL PROTECTED]> > >> Date: 18/08/08 17:18 > >> Subj: Bacula 2.2.8, dbcheck never completes > >> > >> And I and Ryan Novosielski use DBMS MySQL. I empirically determined > >> that the need to create an index > >> CREATE INDEX idxFJ ON File (FileId, JobId); > >> > >> After that dbcheck works VERY quickly. > >> I described the decision [Bacula-users] and [Bacula-devels]. > >> > >> My proposals were as follows: > >> *** include an appropriate comment in the file > >> src/cats/make_mysql_tables.in after rows > >> CREATE TABLE File... > >> (They already have a similar comment.) > >> > >> I received a reply - no. > > > > I don't see that there was any need for a reply. You know the solution > > for you. Create the index, or write a script that creates it, runs > > dbcheck and then removes it. > > > > As we previously explained, items such as indexes are up to the user to > > manage in general. If someone submits a patch to dbcheck that improves > > it and makes it automatic, all the better. > > > >> *** Describe the problem and solution in the documentation, for > >> example, in section Troubleshooting. > >> > >> At this suggestion, I also did not receive the normal response. > >> > >> *** I studied the source code dbcheck.c and concluded that there can > >> not insert the code, which will depend on the particular DBMS (in this > >> case - MySQL). > >> > >> Understand ? > > > > No. > > > >> Conclusion: there is a problem, but to eliminate it, no one will be? > >> Let the problem exists and further? Or how? > > > > This is an Open Source program. You know your solution -- create an > > index. > > > > Most of the users of Bacula are hard working sysadmins with many of their > > own priorities. They are not there to actively work on Bacula but do help > > users find solutions to problems. > > > > Some of us like myself work full time for *free* on projects that the > > users have voted to be the most important. Most like myself are > > generally totally overloaded. We are not complaining, but for programs > > such as dbcheck that are not at all critical and should not be needed to > > run more than once a year, we can only offer advice, which is what we > > have done. > > > > At least it is free and you have the source code, and you have received a > > good deal of feedback. > > > > Best regards, > > > > Kern > > > >> Thanks. > >> > >> 2008/9/3 Kern Sibbald <[EMAIL PROTECTED]>: > >> > On Wednesday 03 September 2008 08:22:31 Yuri Timofeev wrote: > >> >> Hi > >> >> > >> >> 2008/8/31 Eric Bollengier <[EMAIL PROTECTED]>: > >> >> > If you want to use new indexes with dbcheck, you can modify > >> >> > dbcheck.c in this way : > >> >> > - ask to the user if he wants to add them (think about disk space) > >> >> > - add them + run analyse > >> >> > - run the cleanup operation > >> >> > - remove them > >> >> > > >> >> > FYI, postgresql can use composite indexes, so it will require less > >> >> > indexes than mysql. > >> >> > >> >> It seems the problem appears only when working with MYSQL. > >> >> > >> >> I made a small program in order to make temporary index > >> >> CREATE INDEX idxFJ ON File (FileId, JobId); > >> >> and then removes him. > >> >> > >> >> Now I little understand a Bacula API. And there is one problem: > >> >> my source code depends on the DBMS (MySQL). > >> >> And I still do not know how it can be inserted in dbcheck.c > >> >> > >> >> > >> >> If I write (in dbcheck.c) as follows: > >> >> > >> >> # ifdef HAVE_MYSQL > >> >> / * My code uses bacula api * / > >> >> # endif > >> >> > >> >> So it would be wrong, imho, or is this possible? > >> >> Can you suggest any solution? > >> > > >> > Please be specific about the exact MySQL dependencies. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Bacula-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/bacula-devel
